GET (version 0.1-3)

GET.variogram: Variogram and residual variogram with global envelopes

Description

The function accompanies the function variogram with global envelopes that are based on permutations of the variable(s) or residuals for which the variogram is calculated. Therefore, one can inspect the hypothesis of "no spatial autocorrelation" of the variable or the residuals of the fitted model.

Usage

GET.variogram(object, nsim = 999, data = NULL, ..., GET.args = NULL,
  savefuns = TRUE)

Arguments

object

An object of class gstat or a variogram.formula. In the first case, direct (residual) variograms are calculated for the variable defined in object. Only one variable allowed. In the second case, a formula defining the response vector and (possible) regressors, in case of absence of regressors, use e.g. z~1. See variogram.

nsim

The number of permutations.

data

A data frame where the names in formula are to be found. If NULL, the data are assumed to be found in the object.

...

Additional parameters to be passed to variogram.

GET.args

A named list of additional arguments to be passed to global_envelope_test.

savefuns

Logical. If TRUE, then the functions from permutations are saved to the attribute simfuns.

Examples

# NOT RUN {
if(require("sp", quietly=TRUE) & require("gstat", quietly=TRUE)) {
  # Examples from gstat complemented with global envelopes
  #-------------------------------------------------------
  data(meuse)
  coordinates(meuse) <- ~x+y
  # topsoil zinc concentration, mg kg-1 soil ("ppm")
  bubble(meuse, "zinc",
         col=c("#00ff0088", "#00ff0088"), main="zinc concentrations (ppm)")
  # Variogram can be calculated as follows by the function variogram of the gstat library.
  # The function variogram takes a formula as its first argument:
  # log(zinc)~1 means that we assume a constant trend for the variable log(zinc).
  lzn.vgm <- variogram(object=log(zinc)~1, data=meuse)
  plot(lzn.vgm)
  # Variogram with global envelopes is as easy:
  
# }
# NOT RUN {
lzn.vgm.GET <- GET.variogram(object=log(zinc)~1, data=meuse)
# }
# NOT RUN {
  
# }
# NOT RUN {
  plot(lzn.vgm.GET)

  # Instead of the constant mean, denoted by ~1, a mean function can
  # be specified, e.g. using ~sqrt(dist) as a predictor variable:
  lznr.vgm <- variogram(log(zinc)~sqrt(dist), meuse)
  # In this case, the variogram of residuals with respect
  # to a fitted mean function are shown.
  plot(lznr.vgm)
  # The variogram with global envelopes (obtained by permuting the residuals):
  
# }
# NOT RUN {
lznr.vgm.GET <- GET.variogram(object=log(zinc)~sqrt(dist), data=meuse)
# }
# NOT RUN {
  
# }
# NOT RUN {
  plot(lznr.vgm.GET)

  # Directional variograms
  lzn.dir <- variogram(object=log(zinc)~1, data=meuse, alpha=c(0, 45, 90, 135))
  plot(lzn.dir)
  # with global envelopes
  
# }
# NOT RUN {
lzn.dir.GET <- GET.variogram(object=log(zinc)~1, data=meuse, alpha=c(0, 45, 90, 135))
# }
# NOT RUN {
  
# }
# NOT RUN {
  plot(lzn.dir.GET, base_size=10)

  # Use instead gstat objects
  g <- gstat(id="ln.zinc", formula=log(zinc)~1, data=meuse)
  # or: g <- gstat(id="ln.zinc", formula=log(zinc)~sqrt(dist), data=meuse)
  # The variogram
  plot(variogram(g))
  # The variogram with global envelopes:
  
# }
# NOT RUN {
g.GET <- GET.variogram(object=g)
# }
# NOT RUN {
  
# }
# NOT RUN {
  plot(g.GET)
}
# }